Identification and quantification forest 
degradation drivers in tropical dry 
forests: a case study in Western Mexico 
- 
Technical session : “From understanding drivers to gaining leverage 
at the tropical forest margins: 20 years of ASB Partnership” 
IUFRO XXIV World Congress, Salt Lake City, 5 -11 Oct 2014 
Lucia Morales-Barquero, Armonia Borrego, Margaret Skutsch, 
Christoph Kleinn & John Healey.
Contents 
• Introduction 
• Methods 
• Research Approach 
• Results & Discussion 
• Conclusions and Significance
Introduction 
• Tropical Dry Forests (TDFs) and REDD+: 
– TDFs have been neglected in REDD+ because of low 
carbon content and small annual increments 
– Nonetheless TDFs cover approx 42% of tropical and 
subtropical forest areas 
– Many are quite degraded 
– Hold relatively high population 
densities of forest depenent 
people
Introduction 
• TDFs and forest degradation processes: 
– Commonly used for shifting cultivation 
– Source of fence posts, fuelwood, and cattle grazing 
within the forest. 
• Shifting cultivation produces complex landscapes 
in which old forests, forests at various stages of 
secondary growth and agriculture co-exist  
mosaic of patches that are losing or gaining forest 
carbon stocks 
• Should be regarded under REDD+ as leading to 
degradation not deforestation
Introduction 
• In Mexico shifting cultivation (´milpa´ or ´coamil´) is practiced 
by individuals within communal land systems managed by 
Ejidos. Typically 2 years cultivation/5-10 years fallow. Fallow 
provides grazing, poles, and firewood. 
• Ejidos are the target group of REDD+ policies
Introduction 
• Shifting cultivation systems are highly dynamic 
which makes it difficult to assess its changes in 
carbon stocks and the associated causes of those 
changes. 
• Adequate scale , both social and biophysical factors 
is required. 
• Research Objective: 
– To present analytical framework capable of identifying the 
complex of drivers of forest degradation in TDF by using 
satellite images that provide data at a scale fine enough 
to detect forest degradation due to shifting cultivation, 
with on-the-ground survey data on the local use of forest 
resources.
Methods: Study Site 
• Ayuquila Watershed, Jalisco, Pacific Coast 
• REDD+ 
Early Action, 
from Mexican 
Government 
• Organized 
Intermunicipial 
Government
Typical Mosaic Landscape , under different stages of regeneration and clearance. 
Clearings for Shifting Cultivation 
Regrowth TDF after shifting cultivation 
Permanent Agriculture
Research Approach 
Two components to assess the drivers: 
Production of map to assess shifting cultivation at the 
landscape and community level: The method 
developed eliminates permanent agriculture from 
consideration and focuses only on forest cover 
changes (canopy cover) assoicated with shifting 
cultivation. For this, SPOT 5 image data (10x10 m) for 
2004-2010 were used to map small clearings. 
Statistical Model (binary logistic regression) using the 
map as the dependent variable to assess the factors 
that can be associated with the probabiligy of forest 
degradation in shifting cultivation landscapes.
Methods & Materials 
• Model of the 
Probability of 
Forest degradation 
Biophysical 
Variables 
Socio 
Economic 
Variables 
Forest Cover 
Changes 
Binary 
Variable 
Local Surveys 
Semi structure 
Interviews in 29 
Ejidos 
Remote sensing 
Data & Ancillary 
Spatial Data 
•Ratio TDF: 
•Population 
•Livestocks 
•Fence Posts 
•Marginalization 
Index 
•Parcel Size 
•Elevation 
•Slope 
•Distance to 
Main Towns
Results & Discussion 
• Both at landscape and community level clearance and 
regrowth are balanced in area, and our measurements 
suggest that although carbon stocks are on average lower 
than in ´intact´ TDF, shifting cultivation does not cause net 
emissions. Regrowth rates are very rapid in first 5 years and 
soil carbon levels increase. 
Forest Cover Change Map (2004-2010) and summary of the landcover 
transitions
Results & Discussion 
Binary Logistic Regression Model for the Probability of Forest Degradation 
associated with Shifting Cultivation 
Variable Name Estimated 
coefficient (b ) S.E. 
Marginal 
effect 
Slope -0.0545*** 0.0111 -0.5602 
Dist -0.0326* 0.0163 -0.3352 
Pop:TDF -0.0157* 0.0066 -0.1614 
Fence 0.0003*** 0.00008 0.0031 
Livestock 0.0002** 0.00006 0.0021 
HMI 0.7051*** 0.2078 7.2472 
MMI 0.48360* 0.2455 4.9706 
Parcel_T -0.00037* 0.0002 -0.004 
Slope_Elev 0.00003*** 0.00001 0.0003 
Constant -1.495*** 0.6880 -17.051 
n = 1952, S.E. = standard error of estimation of the model, model log likelihood ratio 
= -769.39 (df = 10); AUC = 65.1; residual deviance = 1538.7; null deviance = 1605.2. 
AIC = 1558.78 
The model predicts changes 
from TDF to shifting cultivation 
and vice versa correctly in 65% 
of the cases
Results & Discussion 
• Flatter areas have a higher probablity of being used for 
shifting cultivation (For every 1% increase in slope there is 
0.75% decrease on the probability of forest degradation). 
Risk is especially high in flat areas on hill tops. 
• Ejidos characterized by lower incomes and less education 
are more dependent on shifting cultivation, hence highly 
marginalized communities and medium marginalized 
communities have a greater probability of forest 
degradation (7.25 %, 5% respectively) than communities 
with a low index of marginalization.
Results & Discussion 
• Livestock management and use of fencing posts 
by the ejidos are associated with increases in 
probability of forest degradation. An increase of 
100 posts will mean an increase of 16% in the 
probability of forest degradation 
• Overall biophysical variables (measured at a pixel 
level) contributed 43.7% of the log-likelihood and 
community-level information explained around 
56.3%.
Conclusions and Significance for REDD+ 
• Shifting cultivation over last 50 years has resulted 
in lowered carbon stocks in TDF but is not a net 
emitter of carbon now in the area studied. 
• The potential under REDD+ can only therefore be 
in forest enhancement; strategies which increase 
the forest stock; possibilities include 
– Banning shifting cultivation 
– Increasing length of fallows
Conclusions and Significance for REDD+ 
• Banning shifting cultivation and e.g. pay PES 
instead 
– Socially undesirable (safety net of poorest people) 
– PES payments based on carbon value would not cover 
the opportunity costs 
– Our other studies show that production of equivalent 
quantities of maize in permanent agricultural systems 
would emit MORE carbon
Conclusions and Significance for REDD+ 
• Increasing length of fallow 
– Evidence that fallows have been artificially shortened 
in response to government subsidies (PROCAMPO) 
– Possibility to lengthen fallows limited by land 
availability/pop density 
– Alternative may be to regulate grazing. The impact of 
grazing cattle in fallow recovery needs more study.
Thank you! 
Lucia Morales 
moralesluciacr@gmail.com 
Margaret Skutsch 
mskutsch@ciga.unam.mx

Identification and quantification forest degradation drivers in tropical dry forests: a case study in Western Mexico

  • 1.
    Identification and quantificationforest degradation drivers in tropical dry forests: a case study in Western Mexico - Technical session : “From understanding drivers to gaining leverage at the tropical forest margins: 20 years of ASB Partnership” IUFRO XXIV World Congress, Salt Lake City, 5 -11 Oct 2014 Lucia Morales-Barquero, Armonia Borrego, Margaret Skutsch, Christoph Kleinn & John Healey.
  • 2.
    Contents • Introduction • Methods • Research Approach • Results & Discussion • Conclusions and Significance
  • 3.
    Introduction • TropicalDry Forests (TDFs) and REDD+: – TDFs have been neglected in REDD+ because of low carbon content and small annual increments – Nonetheless TDFs cover approx 42% of tropical and subtropical forest areas – Many are quite degraded – Hold relatively high population densities of forest depenent people
  • 4.
    Introduction • TDFsand forest degradation processes: – Commonly used for shifting cultivation – Source of fence posts, fuelwood, and cattle grazing within the forest. • Shifting cultivation produces complex landscapes in which old forests, forests at various stages of secondary growth and agriculture co-exist  mosaic of patches that are losing or gaining forest carbon stocks • Should be regarded under REDD+ as leading to degradation not deforestation
  • 5.
    Introduction • InMexico shifting cultivation (´milpa´ or ´coamil´) is practiced by individuals within communal land systems managed by Ejidos. Typically 2 years cultivation/5-10 years fallow. Fallow provides grazing, poles, and firewood. • Ejidos are the target group of REDD+ policies
  • 6.
    Introduction • Shiftingcultivation systems are highly dynamic which makes it difficult to assess its changes in carbon stocks and the associated causes of those changes. • Adequate scale , both social and biophysical factors is required. • Research Objective: – To present analytical framework capable of identifying the complex of drivers of forest degradation in TDF by using satellite images that provide data at a scale fine enough to detect forest degradation due to shifting cultivation, with on-the-ground survey data on the local use of forest resources.
  • 7.
    Methods: Study Site • Ayuquila Watershed, Jalisco, Pacific Coast • REDD+ Early Action, from Mexican Government • Organized Intermunicipial Government
  • 8.
    Typical Mosaic Landscape, under different stages of regeneration and clearance. Clearings for Shifting Cultivation Regrowth TDF after shifting cultivation Permanent Agriculture
  • 9.
    Research Approach Twocomponents to assess the drivers: Production of map to assess shifting cultivation at the landscape and community level: The method developed eliminates permanent agriculture from consideration and focuses only on forest cover changes (canopy cover) assoicated with shifting cultivation. For this, SPOT 5 image data (10x10 m) for 2004-2010 were used to map small clearings. Statistical Model (binary logistic regression) using the map as the dependent variable to assess the factors that can be associated with the probabiligy of forest degradation in shifting cultivation landscapes.
  • 10.
    Methods & Materials • Model of the Probability of Forest degradation Biophysical Variables Socio Economic Variables Forest Cover Changes Binary Variable Local Surveys Semi structure Interviews in 29 Ejidos Remote sensing Data & Ancillary Spatial Data •Ratio TDF: •Population •Livestocks •Fence Posts •Marginalization Index •Parcel Size •Elevation •Slope •Distance to Main Towns
  • 11.
    Results & Discussion • Both at landscape and community level clearance and regrowth are balanced in area, and our measurements suggest that although carbon stocks are on average lower than in ´intact´ TDF, shifting cultivation does not cause net emissions. Regrowth rates are very rapid in first 5 years and soil carbon levels increase. Forest Cover Change Map (2004-2010) and summary of the landcover transitions
  • 12.
    Results & Discussion Binary Logistic Regression Model for the Probability of Forest Degradation associated with Shifting Cultivation Variable Name Estimated coefficient (b ) S.E. Marginal effect Slope -0.0545*** 0.0111 -0.5602 Dist -0.0326* 0.0163 -0.3352 Pop:TDF -0.0157* 0.0066 -0.1614 Fence 0.0003*** 0.00008 0.0031 Livestock 0.0002** 0.00006 0.0021 HMI 0.7051*** 0.2078 7.2472 MMI 0.48360* 0.2455 4.9706 Parcel_T -0.00037* 0.0002 -0.004 Slope_Elev 0.00003*** 0.00001 0.0003 Constant -1.495*** 0.6880 -17.051 n = 1952, S.E. = standard error of estimation of the model, model log likelihood ratio = -769.39 (df = 10); AUC = 65.1; residual deviance = 1538.7; null deviance = 1605.2. AIC = 1558.78 The model predicts changes from TDF to shifting cultivation and vice versa correctly in 65% of the cases
  • 13.
    Results & Discussion • Flatter areas have a higher probablity of being used for shifting cultivation (For every 1% increase in slope there is 0.75% decrease on the probability of forest degradation). Risk is especially high in flat areas on hill tops. • Ejidos characterized by lower incomes and less education are more dependent on shifting cultivation, hence highly marginalized communities and medium marginalized communities have a greater probability of forest degradation (7.25 %, 5% respectively) than communities with a low index of marginalization.
  • 14.
    Results & Discussion • Livestock management and use of fencing posts by the ejidos are associated with increases in probability of forest degradation. An increase of 100 posts will mean an increase of 16% in the probability of forest degradation • Overall biophysical variables (measured at a pixel level) contributed 43.7% of the log-likelihood and community-level information explained around 56.3%.
  • 15.
    Conclusions and Significancefor REDD+ • Shifting cultivation over last 50 years has resulted in lowered carbon stocks in TDF but is not a net emitter of carbon now in the area studied. • The potential under REDD+ can only therefore be in forest enhancement; strategies which increase the forest stock; possibilities include – Banning shifting cultivation – Increasing length of fallows
  • 16.
    Conclusions and Significancefor REDD+ • Banning shifting cultivation and e.g. pay PES instead – Socially undesirable (safety net of poorest people) – PES payments based on carbon value would not cover the opportunity costs – Our other studies show that production of equivalent quantities of maize in permanent agricultural systems would emit MORE carbon
  • 17.
    Conclusions and Significancefor REDD+ • Increasing length of fallow – Evidence that fallows have been artificially shortened in response to government subsidies (PROCAMPO) – Possibility to lengthen fallows limited by land availability/pop density – Alternative may be to regulate grazing. The impact of grazing cattle in fallow recovery needs more study.
  • 18.
    Thank you! LuciaMorales moralesluciacr@gmail.com Margaret Skutsch mskutsch@ciga.unam.mx

Editor's Notes

  • #4 Tropical dry forests have not received as much attention as humid forests in the context of REDD+, mainly because they have lower carbon stocks and increments per area (Blackie et al., 2014).
  • #5 Here we used the term shifting cultivation, slash-burn agriculture or swidden cultivation interchangeable. It is important to stress that in our analysis shifting cultivation is understood to cause forest degradation rather than deforestation as such because its cycle of operation involves clearance followed by regrowth, while deforestation implies a permanent change of land use from forest to non-forest. It usually results in the formation of complex mosaic landscapes, in which old forests, forests at various stages of secondary growth and agriculture co-exist in close proximity (ITTO, 2002). As a result, landscapes where shifting cultivation is practiced, are a mosaic of patches that are losing or gaining forest carbon stocks (Mertz et al., 2012). Howdver, although there can be carbon gains at the landscape level during particular periods of time, the resulting secondary forests on average usually hold less carbon stocks than mature TDF (Becknell, Kissing Kucek, & Powers, 2012; Lawrence, Suma, & Mogea, 2005; Read & Lawrence, 2003) and therefore must be considered as degraded forests in the REDD+ context, both in terms of carbon and in their ecological characteristics.
  • #6 Brief description on the system. Important to tell that fallow areas are the source of fence posts, fuelwood and that cattle graze within the regenerated fallows
  • #7 Usually the analysis of forest cover change are done at scales that are far to coarse to detect changes of forest cover cause by shifting cultivation. Likewise, the analysis of proximate causes use readily available data usually at a national scale, that are not appropiate to detect community processess Conclusion of this part difficult systems to assess biomass loss for REDD+ they are complex systems both in social and biophysical aspects.
  • #8 Description of the Ayuquila Project The majority of the area is TDFs which belongs to ejidos. We use a sample of 29 ejidos.
  • #9 Here, we can see the effect of this shifting cultivation cycle, how is it reflected in the landscape
  • #10 The first part of the research involved the production of a TDF cover change map due to shifting cultivation. Forest degradation associated with shifting cultivation is difficult to follow because it is a highly dynamic system in terms of space and time. Previously resolution, in our case we mapped the small clearings that can be associated with shifting cultivation with high certainty, because we used high resolution imagesand a robust classifier. Next, the information derived from the interpretation of this map was used as an input to a statistical model that allows the identification and quantification of the probability of forest degradation from an integrated set of biophysical factors and socio-economic input variables.
  • #11 We developed a binary logistic regression model. Using as dependent variable forest cover change as a proxy of forest degradation. We sampled 29 ejidos- agrarian communities gathered data on the amount of fence posts, and number of livestocks, the average parcel size, population. Elevation, slope and distance to main towns were extracted form remote sensing data sources. A total of 2000 points were used to developed the statistical model.
  • #12 These fine detailed analysis capture small clearings due to shifting cultivation. Around 24% of the area (which was made up of 20,761 ha of clearance and 23,913 ha of regrowth) can be considered as degraded TDF, for the period of time studied. This raise important quesitons on the amount of carbon stocks at the landscape level, there is no net loss.
  • #13 Several models were developed using the variables previously described. Here is the final model. Both biophysical and socioeconomic variables were significantly associated with the probability of TDF degradation . This model could predict correctly 65% of the cases the change from TDF to shifting cultivation.
  • #14 The factors that were the most important in the model were slope and marginalization index. Both biophysical and socioeconomic variables were significantly associated with the probability of TDF degradation (Table 2). The model results indicated that for every 1% increase in slope there is a decrease of 0.75% in the probability of forest degradation and that the slope is the most important factor for determining if an area will be used for shifting cultivation. Risk is especially high in flat areas on hill tops, because there is an interaction between slope and elevation.
  • #16 Two proposed activities to halt forest degradation and increase carbon sequestration are to increase the fallow periods and/or stop shifting cultivation practices to enable TDF to recuperate to its intact state and increase overall carbon stocks. However, designing payments systems for REDD+ to compensate for avoiding degradation by removing shifting cultivation are likely to run into problems in fulfilling the criteria of equity (and in minimizing the risk of “leakage”); unless they are well designed they risk removing the source of livelihood of the most vulnerable community members without adequate compensation, especially in highly marginalized ejidos.
  • #17 Two proposed activities to halt forest degradation and increase carbon sequestration are to increase the fallow periods and/or stop shifting cultivation practices to enable TDF to recuperate to its intact state and increase overall carbon stocks. However, designing payments systems for REDD+ to compensate for avoiding degradation by removing shifting cultivation are likely to run into problems in fulfilling the criteria of equity (and in minimizing the risk of “leakage”); unless they are well designed they risk removing the source of livelihood of the most vulnerable community members without adequate compensation, especially in highly marginalized ejidos.
  • #18 Two proposed activities to halt forest degradation and increase carbon sequestration are to increase the fallow periods and/or stop shifting cultivation practices to enable TDF to recuperate to its intact state and increase overall carbon stocks. However, designing payments systems for REDD+ to compensate for avoiding degradation by removing shifting cultivation are likely to run into problems in fulfilling the criteria of equity (and in minimizing the risk of “leakage”); unless they are well designed they risk removing the source of livelihood of the most vulnerable community members without adequate compensation, especially in highly marginalized ejidos.